This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: PR 30437 [4.0/4.1/4.2/4.3 Regression] -Wno-all is rejected (try 2)
On Fri, Jan 26, 2007 at 12:20:49AM +0000, Manuel L?pez-Ib??ez wrote:
> On 26/01/07, Steve Kargl <sgk@troutmask.apl.washington.edu> wrote:
> >On Thu, Jan 25, 2007 at 11:58:50PM +0000, Manuel L?pez-Ib??ez wrote:
> >> On 25/01/07, Steve Kargl <sgk@troutmask.apl.washington.edu> wrote:
> >>
> >> >You've completely missed the point of my comment. The proper
> >> >fix is to reject -Wno-all in every frontend. Unfortunately,
> >> >your gfortran patch will probably need to stay for consistency
> >> >with all of the other broken frontends.
> >>
> >> You can still reject -Wno-all in gfortran without breaking other
> >> front-ends. Just give an error if value==0. I will prepare the patch
> >> if that makes you happy.
> >
> >You need to leave the patch in for consistency with the other
> >broken frontends.
> >
>
> You could reject -Wno-all without breaking the other front-ends but
> that is not ok. You want now consistency. So breaking the other
> front-ends was intentional or what?
The earliest version of fortran/lang.opt shown in
cvsweb is dated 13-May-04. It shows the RejectNegative on Wall and I
suspect this goes all the way back to 2002. It took at least 3 years
and may be up to 5 years for someone to notice the gfortran's lang.opt
broke the other frontends. I think you can conclude that no one in over
5 year has use -Wno-all. Kinda suggests that -Wno-all is just contributing
to the proliferation of gcc options.
> OK. I give up. You say that the patch stays. So be it. I am not sure
It has to stay because someone is likely to set a Makefile variable
via
CFLAGS = -O2 ... -Wno-all
FFLAGS = ${CFLAGS}
> OK. I give up. You say that the patch stays. So be it. I am not sure
> what we are discussing then. I hope this conversation finishes here
> also for everybody else.
We were discussing whether -Wno-all should be globally rejected,
which is what I think is the most logical option.
--
Steve